Friday, October 11, 2013

Predicting Violent Behavior

The New York Times recently published an article on our (in)ability to predict violent behavior, such as the shooting spree at the Washington Navy Yard.

In the social and behavioral sciences, the word "prediction" has at least two different meanings.  In one sense, we can predict future behavior to the extent we can influence it.  This is the type of prediction involved in applied work.  We call a treatment or other method of behavior change "reliable" and "effective" to the extent that we are confident in the outcome it will produce.  Along these lines, the New York Times suggested a "mental health ambulance" of sorts in which mental health professionals would be dispatched to a scene in order to diffuse the situation.

A second type of prediction  relates to the prediction of events as they occur in naturalistic settings, separate from any treatment or intervention.  A common term for this type of prediction are "forecasting." Weather forecasting aside, the most common type of forecasting we hear about are economic forecasts.  The latter exert significant influence over individuals' monetary decisions.  

Macro-economics aside, behavioral or psychological forecasting is actually more common than you might think.  The Guardian published an article a few years back titled "Forecasting Human Behavior Carries Big Risks" which stated the following: 
The personalised recommendations and special offers that pop up when you order books or groceries online, and even the specific sequence of questions an insurance call centre asks about your claim, are all generated by computerised predictive algorithms derived from analysing patterns, links and associations in large sets of data.

The suggested movies and shows given to you by Netflix, for example, can be said to use a type of behavioral forecasting.  

Both articles suggest that while we may be able to predict large-scale sociological trends over time, the ability to predict what an individual will do in a naturalistic environment, particularly low-rate once-in-a-lifetime behavior, is extremely difficult.  All sciences have their boundaries of knowledge, and this is certainly one for behavioral science.  

However, crossing this boundary may present a host of ethical concerns (e.g., a surveillance state) that mankind simply does not want to deal with.

1 comment:

  1. Some very interesting thoughts here. Where do you draw the line between having a surveillance state and protecting innocent people?

    ReplyDelete